so2jia / oiwe Goto Github PK

#Interpretable Word Embeddings

##Overview This tool provide a method of interpretable word embeddings, which is based on OIWE-IPG model in our paper, Online Learning of Interpretable Word Embeddings.

##Usage

####In Directory word2nve-c/: Use command "make" to compile sourse files in directory. OIWE model is realized in word2nvec.c

####In Directory bin/: The command to run OIWE model is in launcher.sh. Configurations in launcher.sh can be modified. Use "sh launcher.sh" to run the model.

format: ./word2vec -train -output -size -threads -binary <save as binary: 0 or 1> -cbow 0

##Experiments

####Data

• text8 corpus from word2vec website: https://code.google.com/p/word2vec/
• Test datasets on http://wordvectors.org

####Word Similarty Task

####Spearman Coefficient - dimension

####Word Intrusion